- The Journal of Arthroscopic and Related Surgery

Original Article With Video Illustration
Traumatic Glenohumeral Bone Defects and Their Relationship
to Failure of Arthroscopic Bankart Repairs: Significance
of the Inverted-Pear Glenoid and the Humeral
Engaging Hill-Sachs Lesion
Stephen S. Burkhart, M.D., and Joe F. De Beer, M.D.
Purpose: Our goal was to analyze the results of 194 consecutive arthroscopic Bankart repairs (performed by 2 surgeons with
an identical suture anchor technique) in order to identify specific factors related to recurrence of instability. Type of Study:
Case series. Materials and Methods: We analyzed 194 consecutive arthroscopic Bankart repairs by suture anchor technique
performed for traumatic anterior-inferior instability. The average follow-up was 27 months (range, 14 to 79 months). There
were 101 contact athletes (96 South African rugby players and 5 American football players). We identified significant bone
defects on either the humerus or the glenoid as (1) “inverted-pear” glenoid, in which the normally pear-shaped glenoid had
lost enough anterior-inferior bone to assume the shape of an inverted pear; or (2) “engaging” Hill-Sachs lesion of the humerus,
in which the orientation of the Hill-Sachs lesion was such that it engaged the anterior glenoid with the shoulder in abduction
and external rotation. Results: There were 21 recurrent dislocations and subluxations (14 dislocations, 7 subluxations). Of
those 21 shoulders with recurrent instability, 14 had significant bone defects (3 engaging Hill-Sachs and 11 inverted-pear
Bankart lesions). For the group of patients without significant bone defects (173 shoulders), there were 7 recurrences (4%
recurrence rate). For the group with significant bone defects (21 patients), there were 14 recurrences (67% recurrence rate).
For contact athletes without significant bone defects, there was a 6.5% recurrence rate, whereas for contact athletes with
significant bone defects, there was an 89% recurrence rate. Conclusions: (1) Arthroscopic Bankart repairs give results equal
to open Bankart repairs if there are no significant structural bone deficits (engaging Hill-Sachs or inverted-pear Bankart
lesions). (2) Patients with significant bone deficits as defined in this study are not candidates for arthroscopic Bankart repair.
(3) Contact athletes without structural bone deficits may be treated by arthroscopic Bankart repair. However, contact athletes
with bone deficiency require open surgery aimed at their specific anatomic deficiencies. (4) For patients with significant
glenoid bone loss, the surgeon should consider reconstruction by means of the Latarjet procedure, using a large coracoid bone
graft. Key Words: Instability—Arthroscopic instability repair—Shoulder instability—Bone defect—Bone graft—Latarjet
reconstruction.
From the Department of Orthopaedic Surgery, Baylor College of
Medicine and the University of Texas Health Science Center at San
Antonio, San Antonio, Texas, U.S.A. (S.S.B.); and the Sports Science Institute of South Africa, Oranjezicht, Capetown, South Africa
(J.F.DeB.).
Address correspondence and reprint requests to Stephen S.
Burkhart, M.D., 540 Madison Oak Dr, Suite 620, San Antonio, TX
78258, U.S.A.
© 2000 by the Arthroscopy Association of North America
0749-8063/00/1607-2511$3.00/0
doi: 10.1053/jars.2000.17715
NOTE: To access the video illustration accompanying this
report, visit the October on-line issue of Arthroscopy at www.
arthroscopyjournal.org
T
he debate over the supremacy of open versus
arthroscopic surgical repair for traumatic anterior
instability rages now more energetically than ever.
After more than a decade of unfocused confrontation,
the debate has finally crystallized into a classic conflict between “lumpers” (the open proponents) and
“splitters” (the arthroscopic proponents). On the one
hand, the lumpers view every report of high recurrence rates from outdated transglenoid repairs as evidence that all arthroscopic repairs are somehow inherently inferior to the “gold standard” open repairs. On
the other hand, the arthroscopists, who have evolved
Arthroscopy: The Journal of Arthroscopic and Related Surgery, Vol 16, No 7 (October), 2000: pp 677– 694
677
678
S. S. BURKHART AND J. F. DE BEER
into splitters, acknowledge that there have been problems with prior arthroscopic techniques but insist that
they have now identified the essential surgical element
that will produce results as good as the open gold
standard.
Unfortunately for the arthroscopist, the issue is not
so simple as using a single elusive technical innovation that will assure success in each case. Such a
simplistic approach might be satisfactory for a handyman, but not for a scientist. The arthroscopic surgeon
must recognize the pathology in each case and incorporate techniques that restore both the anatomy and
the biomechanical function of the damaged structures.
You would think we would have made what we feel
is our most critical observation before now. We are,
after all, bone surgeons. What we are referring to is
the fact that most of our arthroscopic Bankart repair
failures have resulted from traumatic bone defects,
either on the glenoid side or on the humeral side. The
underlying cause of our failures has not been inadequate soft tissue fixation, but rather traumatic bone
deficiency. The basic cause of postoperative recurrent
dislocation has less to do with whether the repair is
performed open or arthroscopically than it has to do
with recognition and adequate treatment of mechanically significant bone defects.
At the close of the 1990s, we witnessed a phenomenal amount of effort being directed at shrinkage or
plication of capsular tissues in the treatment of instability. The focus of technical refinements has been
almost entirely on the soft tissues. We strongly believe
that this focus has been misdirected and has diverted
our attention from the main problem: traumatic glenohumeral bone defects. Therefore, we undertook this
study in an attempt to analyze traumatic bone deficiency as a factor in redislocation after Bankart repair.
MATERIALS AND METHODS
We analyzed 194 arthroscopic Bankart repairs performed by the two of us between July 1992 and June
1998. One of us (S.S.B.) performed 43 of these surgeries between July 1992 and June 1998, and the other
(J.F.DeB.) performed 151 of the surgeries between
January 1994 and May 1998. This was a consecutive
series of traumatic anterior instability for each author,
with the following exclusions: anterior instability due
to humeral avulsion of the glenohumeral ligaments
(HAGL lesions) and cases with associated subscapularis tendon avulsions were deemed not to be appropriate for arthroscopic repair. Eight of the patients had
had previous surgery for instability. One author
(S.S.B.) had an average follow-up of 31 months
(range, 14 to 79 months) and the other author had an
average follow-up of 26 months (range, 14 to 54
months). For the combined series of 194 patients, the
average follow-up was 27 months (range, 14 to 79
months). All patients had a minimum follow-up of 14
months.
The average age at the time of surgery was 27.9
years (range, 15 to 64 years). There were 170 male
patients (87.6% of the total) and 24 female patients
(12.4% of the total). The dominant arm was involved
in 141 of the 194 shoulders (72.7%). All patients had
sustained at least 1 traumatic anterior dislocation.
There were 101 contact athletes (96 South African
rugby players and 5 American football players). The
other dislocations were sustained in various noncontact athletic activities or in industrial or household
accidents.
Significant bone defects were either on the humerus
or the glenoid. We defined significant humeral bone
defects to be “engaging” Hill-Sachs lesions. These
lesions, to be explained in greater detail later in this
article, were oriented in such a way that they engaged
the anterior glenoid in a position of athletic function
(defined as 90° abduction combined with external
rotation anywhere between 0° and 135°). We defined
a significant glenoid bone defect as one in which the
arthroscopic appearance of the glenoid (as viewed
through a posterior or anterior portal with the arthroscope placed superiorly and viewing inferiorly) was
that of an “inverted pear.” The normal glenoid is pear
shaped, with a larger diameter below the midglenoid
notch than above it. With the inverted-pear glenoid,
the inferior glenoid had a smaller diameter than the
superior glenoid. There were significant bone defects
in 21 patients (3 engaging Hill-Sachs lesions, and 18
inverted-pear Bankart lesions). Nine of these 21 patients were contact athletes (all rugby players). There
were 16 first-time dislocations, 4 of which had significant bone defects (all were glenoid bone defects).
Both authors performed an identical arthroscopic
Bankart repair using a variety of metallic suture anchors. The glenoid neck was lightly prepared with a
high-speed burr, being careful not to remove significant amounts of bone and thus contribute to bone
deficiency. The capsuloligamentous complex was dissected free of the underlying subscapularis so that it
could be easily reapproximated anatomically to the
“corner” of the glenoid, with suture anchors that were
placed approximately 1 cm apart. Care was taken not
to medialize the suture anchors. Braided polyester
suture was used for the repair. An average of 3 suture
FAILED ARTHROSCOPIC BANKART REPAIR
anchors was used per case. Anatomic reconstruction
was the goal, without any attempt to take up additional
slack in the tissues by techniques such as rotator
interval closure or thermal shrinkage.
Patients with significant bone defects discovered
arthroscopically underwent immediate open surgery.
Those with engaging Hill-Sachs lesions underwent
open capsular shift reconstruction,1 and those with
inverted-pear glenoids had an open Latarjet procedure.2,3 Although these patients were not a part of the
194 patients in this series, their open treatment modalities are outlined in the Discussion for completeness of the treatment protocol.
The postoperative protocol required immobilization
of the operated extremity in a sling or shoulder immobilizer for 3 weeks. Forward flexion was begun 3
weeks postoperatively, and external rotation at 6
weeks. Strengthening exercises were begun 8 weeks
postoperatively.
Statistical analysis was used to compare recurrence
rates for patients with bone deficiency versus those
without bone deficiency. Pearson’s ␹-square test and
the Fisher exact probability test were used.
RESULTS
We had 21 recurrent dislocations and subluxations
(14 dislocations, 7 subluxations), for a 10.8% recurrence rate. However, of those 21 recurrent dislocations
and subluxations, 14 had significant bone defects (3
engaging Hill-Sachs lesions and 11 inverted-pear Bankart lesions).
We found it useful to divide our patients into 2
groups: those without significant bone defects (173
patients) and those with significant bone defects (21
patients). For the group without significant bone defects, there were 7 recurrences in 173 patients, for a
4% recurrence rate. For the group with significant
bone defects, there were14 recurrences in 21 patients,
for a 67% recurrence rate. Thus, the recurrence rate
for patients without significant bone defects was found
to be significantly lower than the recurrence rate for
patients with significant bone defects, with a P value
approaching zero (P ⬍ .0001), when analyzed both by
Pearson’s ␹-square test and the Fisher exact probability test.
Of the 101 contact athletes (96 rugby players and 5
American football players), there were 9 significant
bone defects, all in rugby players, 3 of which were
first-time dislocators. Of the 9 rugby players with
significant bone defects, 8 have redislocated, for an
89% recurrence rate in rugby players with bone de-
679
fects. For rugby players without bone defects (87
patients), there were 6 recurrences, for a 7% recurrence rate. For American football players (5 patients),
none of whom had bone defects, there were no recurrences. The combined recurrence rate for contact athletes (rugby and American football players) without
bone defects was 6 of 92, or 6.5%. This recurrence
rate for contact athletes without significant bone defects is significantly lower than that for contact athletes with significant bone defects, when analyzed by
the aforementioned statistical methods (P ⬍ .0001).
DISCUSSION
Arthroscopic Versus Open Repairs
Arthroscopic Bankart repairs remain controversial
primarily because of reports of high recurrence rates,
ranging up to 44% for transglenoid repairs.4,5 However, on critical analysis of the literature, the recurrence rate is quite variable and can be very low even
with transglenoid sutures.6,7 A report on a recent series of arthroscopic suture anchor repairs8 described
excellent results with only a 7% recurrence rate in an
athletic population. The results in this arthroscopic
suture anchor study equal those of the gold standard
open Bankart repair.9
If one looks at arthroscopic reports, the best results
are those that emphasize repair of the capsulolabral
tissue to the corner of the glenoid, or even onto the
face of the glenoid.6-8,10-12 This should not be surprising, since Neviaser13 described a fairly common pattern of medialized capsulolabral healing (the ALPSA
lesion, or anterior labroligamentous periosteal sleeve
avulsion) associated with recurrent dislocation. If we
repair the capsule in a medialized position, the position of an ALPSA lesion, we would expect a higher
recurrence rate. Indeed, this is exactly where many of
the techniques of arthroscopic staple capsulorraphy or
transglenoid labral repair positioned the labrum, so
one should expect a relatively high failure rate with
this approach.
In contrast, the open Bankart repair as described by
Rowe9 and by Thomas and Matsen14 used transosseous tunnels for suture that exited onto the face of the
glenoid and automatically lateralized the capsulolabral
repair. Suture anchor techniques, both open15 and arthroscopic,6-8,10-12 that lateralize the labrum have also
been shown to give excellent results.
There has been much debate recently in orthopaedic
circles about open versus arthroscopic Bankart repairs
and whether the arthroscopic repairs are inherently
680
S. S. BURKHART AND J. F. DE BEER
weaker for some reason, such as having less inflammatory response with fewer “spot-welds” (suture anchor fixation points). We do not believe that there is
any evidence of inherent weakness in arthroscopic
repairs. One need only look at the results of the open
Dutoit staple capsulorraphy to realize that the issue is
not whether the operation is open or arthroscopic. The
Dutoit procedure16 medialized the capsulolabral complex to create a surgically produced ALPSA lesion.
Not surprisingly, long-term follow-up of this open
stabilization procedure showed an unacceptable recurrence rate of 22%.17 This recurrence rate is in the same
range as that of the arthroscopic staple capsulorraphy,
which Johnson reported to be 21%.18 In this case, the
problem was not that the procedure was done arthroscopically, but that the repair was medialized.
Granted, the untoward effects of medializing the repair had not yet been discovered when either of these
clinical series was performed. However, the fact remains that a bad operation is a bad operation whether
it is performed arthroscopically or open.
For a straightforward Bankart lesion, current reports
indicate that either open or arthroscopic Bankart repair
can give equally good results.6,12,14,15 On the other
hand, an operation whose principles are ill conceived
(e.g., medialization of the capsuloligamentous complex) will have poor results whether it is performed
open or arthroscopic.
Recurrence Related to Glenohumeral
Bone Defects
In our series of arthroscopic Bankart repairs, we
defined a significant glenoid bone defect as one in
which the arthroscopic appearance of the glenoid
when viewed from a superior-to-inferior perspective
was that of an inverted pear (see Video). The geometry of the inverted-pear glenoid is the reverse of the
normal pear-shaped glenoid. This pathologic bonedeficient variation has a smaller diameter below the
midglenoid notch than above it, and this configuration
is obvious when viewed arthroscopically.
On the humeral side, we defined a significant bone
defect to be an engaging Hill-Sachs lesion (see
Video). These defects were Hill-Sachs lesions that we
could see arthroscopically that engaged the anterior
rim of the glenoid when the arm was brought into a
position of athletic function (a position that we defined
as 90° abduction combined with external rotation in
the range between 0° and 135°).
With the above definition of significant glenohumeral bone defects, all 194 of our patients fell clearly
into 1 of 2 categories: those with significant bone
defects and those without them. When we compared
the recurrence rates for these 2 groups of patients,
there was a striking difference: the recurrence rate for
the standard arthroscopic Bankart repair without a
significant bone defect (4% recurrence) was equal to
that of the gold standard open repair, whereas the
recurrence rate for the arthroscopic Bankart repair in
the face of a significant bone defect was an unacceptable 67%. Further subdividing the patients with major
bone defects into glenoid and humeral defects, we
found that those with an inverted-pear glenoid had a
61% (11 of 18) recurrence rate and that those with an
engaging Hill-Sachs lesion had a 100% (3 of 3) recurrence rate.
How can we explain the difference in behavior and
prognosis in these shoulders with significant bone
defects? The answer lies in the reconfigured geometry
of the injured shoulder with a traumatic humeral or
glenoid defect, as we shall discuss.
The Humeral Side: The Hill-Sachs Lesion as a
Cause of Articular-Arc Deficit
The shoulders with large Hill-Sachs lesions that
were rated as failures did not redislocate. In fact, they
probably did not even resubluxate. What these patients complained of was a catching or popping sensation with the shoulder in the abducted and externally
rotated position. They had enough apprehension in
that position that they voluntarily curtailed overhead
activities. Despite the fact that these patients were
probably not true subluxators, they were classified as
failures because their symptoms of instability recurred.
One of us (S.S.B.) had an opportunity to perform a
second-look arthroscopy on 1 of these patients19 and
found that the Bankart lesion had healed. Dynamic
arthroscopic examination of the shoulder as it went
into abduction and external rotation revealed the geometric etiology of the symptoms: there was an articular-arc deficit on the humeral side with an engaging
Hill-Sachs lesion (Fig 1). That is, with the arm in
abduction of 90°, if the shoulder was externally rotated more than 30°, the Hill-Sachs lesion would engage the anterior corner of the glenoid, and the patient
would sense that engagement as a popping or catching
sensation. (See Case Report in this issue: Burkhart and
Danaceau. Articular arc-length mismatch as a cause of
failed Bankart repair.19)
FAILED ARTHROSCOPIC BANKART REPAIR
681
FIGURE 1. This large Hill-Sachs lesion involves a large portion of the humeral articular surface. In this case, even without a Bankart lesion,
the Hill-Sachs lesion can engage the anterior corner of the glenoid, causing symptoms similar to subluxation. We call this an articular arc
deficit.
Engaging Versus Nonengaging Hill-Sachs Lesions
This brings up the issue of engaging versus nonengaging Hill-Sachs lesions. We have not seen this
topic discussed previously in the literature, but we
feel it is very important in predicting the success of
arthroscopic Bankart repairs. We define an engaging Hill-Sachs lesion as one that presents the long
axis of its defect parallel to the anterior glenoid
with the shoulder in a functional position of abduction and external rotation, so that the Hill-Sachs
lesion engages the corner of the glenoid (Fig 2). A
nonengaging Hill-Sachs lesion is one that presents
the long axis of its defect at a diagonal, nonparallel
angle to the anterior glenoid with the shoulder in a
functional position of abduction and external rotation (Fig 3), or one in which the “engagement
point” occurs with the arm in a nonfunctional position of shoulder extension or of low shoulder
abduction (⬍70° abduction). Because this first type
of nonengaging Hill-Sachs lesion passes diagonally
across the anterior glenoid with external rotation,
there is continual contact of the articular surfaces
and nonengagement of the Hill-Sachs lesion by the
anterior glenoid. Such shoulders are reasonable candidates for arthroscopic Bankart repair because they
do not have a functional articular-arc deficit.
Obviously, for every Hill-Sachs lesion, there is a
position of the shoulder at which the humeral bone
defect will engage the anterior glenoid. The symptoms
are greatest if the engagement occurs with the shoulder in a functional position, which typically involves a
combination of flexion, abduction, and external rota-
tion. However, we have found that many Hill-Sachs
lesions engage only when the shoulder is in some
degree of extension, which is a nonfunctional position
for everything except throwing a baseball, or in abduction of less than 70°, which is also a nonfunctional
position. Therefore, we define this second group of
Hill-Sachs lesions as nonengaging.
The orientation of the Hill-Sachs lesion is determined solely by the position of the humeral head
relative to the glenoid when it becomes indented by
the glenoid. This can occur with the shoulder in any
degree of abduction or with the arm at the side and
is not necessarily the degree of abduction in which
the shoulder dislocated. For example, the shoulder
may dislocate with the arm at 90° of abduction, and
then assume a position of 0° abduction after the
dislocation. Hence, the Hill-Sachs lesion that becomes indented with the arm at the side with some
extension of the shoulder will be located more
vertically and superiorly than the lesion that indents
with the shoulder abducted and externally rotated.
This former lesion (the Hill-Sachs that becomes
indented with the arm at the side) is generally a
nonengaging lesion.
Once we recognize an engaging Hill-Sachs lesion
with an articular-arc deficit, we must not only repair the Bankart lesion (if there is one), but we must
also keep the Hill-Sachs lesion from engaging the
anterior glenoid. This can be done in 1 of 3 ways:
First, the surgeon can restrict external rotation
enough that the lesion will not engage (Fig 4). We are
of the opinion that this can be accomplished most
predictably by an open capsular shift procedure.1 We
682
S. S. BURKHART AND J. F. DE BEER
FIGURE 2. Engaging Hill-Sachs lesion. In a functional position of abduction and external rotation, the long axis of the Hill-Sachs
lesion is parallel to the glenoid and engages its anterior corner. (A) Creation of lesion with arm in abduction and external rotation. (B)
Orientation of Hill-Sachs lesion. (C) Engagement of Hill-Sachs lesion in functional position of abduction and external rotation. (D)
Arthroscopic photograph of an engaging Hill-Sachs lesion in a right shoulder as it approaches its engagement position with the anterior
glenoid rim.
do not believe that arthroscopic capsular plication or
thermal capsulorraphy can predictably restrict external
rotation to the extent that may be required, so we
recommend open capsular shift in this situation.
Second, the surgeon may choose not to restrict
external rotation, but instead to lengthen the articular arc of the humerus and fill in the Hill-Sachs
defect so that it can no longer engage. This can be
performed most precisely by a size-matched humeral osteoarticular allograft, but it can also be
accomplished with a corticocancellous iliac graft.20
We recommend this procedure in patients in whom
it is important to maintain external rotation, such as
overhead athletes, or in patients who fail a capsular
FAILED ARTHROSCOPIC BANKART REPAIR
683
FIGURE 3. Nonengaging Hill-Sachs lesion. (A) This Hill-Sachs
lesion was created with the arm at the side and in some extension
and will engage only with the arm at the side with external rotation
and extension, which is not a functional position. (B) Orientation of
Hill-Sachs lesion. (C) In a functional position of abduction and
external rotation, the Hill-Sachs lesion is diagonal to the anterior
margin of the glenoid and does not engage.
shift procedure and begin to engage the Hill-Sachs
lesion again.
Third, the surgeon may try to prevent the HillSachs lesion from engaging by performing a rotational proximal humeral osteotomy21 and internally
rotating the articular surface of the humerus. If the
articular surface is adequately rotated, the HillSachs lesion will not engage. This is a formidable
operation with significant potential morbidity, so
we recommend it only if all other treatment options
have failed.
The Glenoid Side: The Inverted-Pear
Configuration Caused by Bony Bankart or
Impression Bankart Lesions Resulting in
Containment Failure
All of our failures that were associated with bony
Bankart lesions had anterior-inferior glenoid defects
that were large enough to significantly narrow the
inferior half of the glenoid. Ordinarily, the glenoid,
when viewed en face, has the shape of a pear, with the
lower half being significantly wider than the upper
684
S. S. BURKHART AND J. F. DE BEER
FIGURE 4. Capsular plication can restrict external rotation enough
so that the Hill-Sachs lesion will not engage the anterior glenoid.
half (Fig 5A). With a large bony Bankart lesion, or
even a Bankart lesion without an associated bone
fragment but with a significant impression (compression) defect, the shape of the glenoid changes to that
of an inverted pear, where the top half of the glenoid
is wider than the lower half (Figs 5B and C). This
inverted pear is immediately recognizable when the
surgeon views through an anterosuperior portal and
looks down inferiorly on the glenoid. Bigliani et al.22
reported a 12% recurrence rate in patients with glenoid rim fractures who had undergone Bankart repair.
They recommended coracoid transfer if the glenoid
rim fracture comprised 25% of the anterior-posterior
diameter of the glenoid. They did not specify the type
of coracoid transfer (Bristow v the much larger cora-
FIGURE 5. (A) The normal shape of the glenoid is that of a pear, larger below than above. (B) A bony Bankart lesion can create an
inverted-pear configuration. (C) A compression Bankart lesion can also create an inverted pear. (D) Arthroscopic view of an inverted-pear
glenoid (left shoulder) as viewed from an anterosuperior portal. Note how the inferior glenoid (at top) narrows to an apex that is much
narrower than the superior glenoid (at bottom). (E) Left shoulder viewed through an anterosuperior portal. The tip of the hook probe rests
on the bare spot of the glenoid, and the 3-mm laser marks on the probe indicate a 12-mm distance from the bare spot to the posterior glenoid
rim. (F) The probe is placed just inferior to the bare spot, showing that the distance from the bare spot to the anterior glenoid rim is 6 mm,
suggesting a 6-mm bone loss from compression of the anterior glenoid. This represents a 25% reduction in the diameter of the inferior
glenoid.
FAILED ARTHROSCOPIC BANKART REPAIR
coid graft of the Latarjet procedure) that they recommended.
Impression Bankart lesions result from compression
of the anterior glenoid and have no bony fragment to
indicate the amount of bone that has been lost from the
articular arc. We are in the process of completing an
anatomic study that appears to confirm that the bare
spot of the normal glenoid lies equidistant between the
anterior and posterior rims of the glenoid and, thus, is
a good reference point for estimating the percentage of
glenoid bone loss from compression (Figs 5E and F).
In the case of bony Bankart lesions, if one excises
the bony Bankart fragment and repairs the capsulolabral complex to the remaining glenoid despite an
inverted-pear configuration, the repair may seem secure. However, the glenoid bone loss can create a
serious containment problem for the shoulder.
Containment of the humeral head by the glenoid is
a result of 2 geometric variables. The first is the
“deepening effect” of a wider glenoid due to the
longer arc of its concave surface (Fig 6A). This serves
FIGURE 6. (A) The anterior glenoid
rim serves to “deepen the dish” of the
glenoid and acts as a buttress to resist
dislocation. (B) A shoulder with a
bony Bankart lesion has a shallower
“dish” anteriorly with less resistance
to shear forces.
685
to “deepen the dish” of the glenoid. Loss of part of the
glenoid surface will cause the glenoid “dish” to be
shallower and, therefore, less resistant to shear forces
that might tend to cause a dislocation (Fig 6B).
The second geometric variable affecting containment of the humeral head is the arc length of the
glenoid. Axial humeral forces are resisted by the glenoid until the direction of the force vector passes
beyond the edge of the glenoid (Fig 7). At that point,
such forces are concentrated at the bone-ligament
interface and can cause a Bankart lesion. With anterior-inferior glenoid bone deficiency, the safe zone, or
arc through which the glenoid resists axial forces, may
be much smaller than in a normal joint (Figs 8 and 9),
so that the glenohumeral ligaments are presented with
a load that would normally be resisted by the bony
buttress of the glenoid. No wonder the inverted-pear
configuration predisposes to ligament disruption and
recurrent dislocation. Our series confirms the predilection of the inverted-pear configuration for recurrent
dislocation. Of 18 shoulders with the inverted-pear
686
S. S. BURKHART AND J. F. DE BEER
FIGURE 7. Effect of glenoid arc
length. The glenoid resists axial
forces at a variety of angles from the
humerus throughout the glenoid arc
length (A and B). Force vectors that
pass beyond the edge of the glenoid
can create Bankart lesions. Loss of a
part of the glenoid articular surface
(e.g., a bony Bankart lesion) shortens
the arc through which the glenoid
can resist humeral forces.
glenoid, 11 redislocated after arthroscopic Bankart
repair, for a 61% recurrence rate.
Correcting the Inverted Pear: The Latarjet
Procedure
The most logical solution to the problem of anterior
glenoid bone deficiency is to rebuild this bony anterior
buttress with a bone graft. However, free iliac grafts to
the glenoid may not reliably hold up and are prone to
resorption. The Bristow23 procedure has too small a
graft surface to create a bony buttress or to enlarge the
arc length of the glenoid, and it primarily depends on
the sling effect of the conjoined tendon as the shoulder
is brought into abduction and external rotation.
Our preference is to reconstruct the anterior-inferior
glenoid by means of the Latarjet procedure.2,3 This
operation differs from the Bristow procedure in that a
much larger piece of bone (about 2 to 3 cm in length)
is used to create a more formidable arc length exten-
sion of the anterior-inferior glenoid. The procedure is
performed as follows:
As a final check of the anterior rim before committing to an osteotomy of the coracoid, the glenoid is
exposed. This exposure is accomplished by detaching
the upper half of the subscapularis and dissecting it off
the capsule before capsular incision. The inferior portion of the capsule must be dissected from the deep
surface of the lower subscapularis (Fig 10). If this is
not done, the inferior part of the capsule will adhere to
the subscapularis, and the surgeon will have difficulty
repairing it later to the corner of the remaining glenoid. The anterior capsuloligamentous complex is dissected medially as far as possible before transecting
the capsular sleeve to be sure there is sufficient length
to the capsule to later repair to the glenoid rim (Figs
11A and B). This step may also be performed after
coracoid osteotomy when there is better visualization.
The pectoralis minor is dissected from the coracoid
FAILED ARTHROSCOPIC BANKART REPAIR
FIGURE 8. Glenoid bone loss shortens the “safe arc” through
which the glenoid can resist axial forces. Ø2 (bone-deficient condition) is less than Ø1 (normal glenoid).
leaving a sliver of bone with the tendon (Figs 12 and
13), then the coracoid is osteotomized just proximal to
the angle (elbow) of the coracoid, from medial to
lateral, using an osteotome or angled saw (Fig 14).
The osteotomy should traverse the coracoid between
the pectoralis minor insertion and the attachment of
the coracoclavicular ligaments. The coracoclavicular
ligaments must be left intact so as not to destabilize
the acromioclavicular joint. The osteotomy may be
done with an osteotome, but an angled saw is particularly helpful in cutting the thick superolateral cortex
of the coracoid.
The conjoined tendon is left attached to the coracoid
graft because this provides some blood supply to the
coracoid and makes this a vascularized graft; the
transferred coracoid graft can continue to serve as a
stable attachment point for the conjoined tendon. The
coracoid graft is rotated about its long axis to obtain
the best fit against the anterior glenoid, and then a
high-speed burr is used to contour the graft for an
exact fit to the anterior glenoid. At this point, if it has
not already been done, the capsule is dissected from
687
the glenoid as far medially as possible before incising
it for later repair. It is often easier to dissect the
capsule from the glenoid after the coracoid has been
osteotomized. Suture anchors are then placed along
the glenoid at the 3, 4, and 5 o’clock positions for later
capsulolabral repair after the coracoid graft is in place
(Fig 15). The coracoid graft is fixed to the glenoid
with 2 bicortical screws (Fig 16). Then the capsule is
repaired to the glenoid by means of the previously
placed suture anchors (Fig 17A and B). In this way,
the graft becomes an extra-articular platform to extend
the safe arc through which the glenoid can resist axial
forces (Fig 18). It is important to note that the graft is
not intended to be a bone block (Fig 19). The detached
pectoralis minor may be repaired to the base of the
coracoid with a suture anchor (per J.F.DeB.) or simply
approximated to the adjacent soft tissues (per S.S.B.).
Finally, the upper half of the subscapularis muscle is
repaired over the bone graft, with the conjoined tendon penetrating anteriorly between the upper and
lower halves of the subscapularis (Fig 20).
Allain et al.3 recently reported a high rate of late
osteoarthritis following the Latarjet procedure (58%
of 52 shoulders). This was a long-term follow-up
study (average, 14.3 years), and the majority of patients with osteoarthritis had only grade I findings.
The factors that these authors identified as predisposing toward osteoarthritis were (1) coracoid graft placement too far laterally and (2) coexistent rotator cuff
tear at the time of the original injury. One must
wonder whether the late osteoarthritis is due at least
partially to the fact that these individuals had sustained intra-articular fractures predisposing them to a
higher risk of late degenerative change. Interestingly,
none of the 52 shoulders had redislocated at an average of 14.3 years postoperatively, testifying to the
stability of this construct.
In patients who have both an inverted-pear glenoid
and an engaging Hill-Sachs lesion, we have found that
the Latarjet procedure alone is usually adequate to
treat this combined bone deficiency. In this case, the
bone graft corrects the glenoid deficiency so that it can
resist axial forces across an expanded glenoid diameter, and the graft also lengthens the glenoid articular
arc to prevent the Hill-Sachs lesion from engaging.
Contact Athletes
Contact athletes represented a significant segment
of our study population. There were 96 rugby players
and 5 American football players, for a total of 101
contact athletes. The authors have approximately
688
S. S. BURKHART AND J. F. DE BEER
FIGURE 9. (A) Axial force from
the humerus applied centrally on
the glenoid will not create a Bankart lesion or failure of a soft tissue
Bankart repair. (B) If an axial force
is applied through a point beyond
the edge of the deficient glenoid,
failure of a soft-tissue Bankart repair is likely because the load must
be borne by the soft tissues.
equal percentages of their practices devoted to sports
medicine, yet the rugby players represented 63% of
the South African patients in this study whereas the
American football players comprised only 12% of the
American patients. We realize that there may be some
subtle demographic reasons for the disparity in the
percentage of contact athletes operated on by the 2
authors (e.g., referral patterns to each surgeon). Even
so, with the rugby players representing such a significantly higher percentage of the total than the American football players, we have concluded that playing
rugby produces a higher incidence of anterior dislocation than American football. Additionally, the rugby
players had a much higher incidence of significant
bone defects that the American football players (9.4%
for rugby v 0% for American football), implying that
there must be a different mechanism of injury between
the 2 groups. In this regard, we believe that a unique
strength of this study is the fact that it has allowed us
to compare 2 different categories of collision athletes
and search for specific factors that might explain the
differences in outcome between the 2 groups of athletes.
FIGURE 10. Management of the subscapularis: detatch the superior half of the tendon, then develop a plane between the inferior
half of the subscapularis and the capsule.
FAILED ARTHROSCOPIC BANKART REPAIR
689
FIGURE 11. (A) Outline of capsulotomy. (B) Dissect the capsule medially as far as possible before detaching it from the glenoid neck to
preserve as much capsular length as possible for later reattachment.
In examining the mechanism of injury, one must
look at the differences in shoulder position at skill
positions between the 2 sports. In American football,
the running backs generally have their arms held close
to their bodies and seldom sustain dislocations. Many
of the dislocations occur in defensive players making
FIGURE 12. Tendon attachments onto the coracoid before coracoid osteotomy.
arm tackles or sustaining awkward falls to the ground
with the arm in abduction and external rotation. On
the other hand, rugby players who carry the ball in one
FIGURE 13. Pectoralis minor (insertion at dotted line) attachment
is removed from the coracoid so that the osteotomy can be made
proximal to the “elbow” of the coracoid.
690
S. S. BURKHART AND J. F. DE BEER
studied in this series, our data indicate that those
athletes without significant bone defects who have
arthroscopic Bankart repairs are not at any greater risk
than the noncontact athletes for failure of the repair
(P ⫽ .156).
The group of contact athletes particularly at risk of
recurrence is the group with significant bone defects,
which in our study was comprised mainly of rugby
players. In this group, we recommend the Latarjet
reconstruction for inverted-pear glenoid defects, and
we recommend the open capsular shift procedure for
patients with engaging Hill-Sachs lesions.
The Folly of Overconstraining the Soft Tissues
FIGURE 14. Osteotomy of coracoid. Note that the osteotomy is
made proximal to the “elbow” of the coracoid. The osteotomy must
not angle too steeply toward the glenoid, or it may create an
intra-articular glenoid fracture. From a technical standpoint, it is
much easier to make the osteotomy at the proper angle if the
surgeon removes the medial retractor and allows the osteotome
itself to rest on the skin edge and act as its own retractor.
arm will generally “stiff-arm” the opposition with the
other arm. The arm used to stiff-arm the defenders is
the one that is typically dislocated, and this occurs
with an axially directed force while the shoulder is in
about 70° of abduction and 30° of extension. In this
position, the vector of the axially directed force is
brought to bear on the anterior glenoid rim, causing an
intra-articular fracture (bony Bankart lesion) rather
than soft tissue disruption from bone. In contrast, the
American football injuries are generally rotational injuries to the shoulder in which obliquely directed
forces applied to the hand or arm produce significant
moments and torsional effects at the shoulder that tend
to peel the glenohumeral ligaments from the anterior
glenoid.
Comparing the rate of failure of arthroscopic Bankart repairs among the categories of contact athletes,
the most striking finding is the vastly higher failure
rate for those with significant bone defects (89% recurrence rate) compared with those without significant
bone defects (6.5% recurrence rate). The American
football players, none of whom had any significant
bone defects, had no recurrences. This experience
with contact athletes suggests that the conventional
wisdom of making repairs on all contact athletes by
open means is overly simplistic. Based on the athletes
Based on our experience with this group of patients,
we do not understand the current emphasis on greater
soft tissue constraint to “improve” the results of arthroscopic Bankart repairs. The methods may vary
(thermal capsular shrinkage, adjunctive capsular plication, rotator interval closure), but the goal of these
adjunctive procedures is the same—to reduce the recurrence rate by limiting the extremes of motion.
Baker et al.24 have arthroscopically observed intrasubstance capsular tearing without Bankart lesions in
some first-time dislocators. In addition, Bigliani et
al.25 have reported tensile failure of the inferior glenohumeral ligament in which ligament disruption is
preceded by ligament stretching, and they suggest the
possibility of residual laxity due to plastic deformation
resulting from stretching. However, to our knowledge,
there are no clinical or basic science studies proving
FIGURE 15. Suture anchors are placed at the 3, 4, and 5 o’clock
positions for later reattachment of the capsulolabral complex.
FAILED ARTHROSCOPIC BANKART REPAIR
691
FIGURE 16. (A) Coracoid graft is fixed to glenoid with 2 bicortical screws. If need be, the graft can be contoured with a power burr to fit
the curve of the anterior-inferior glenoid. (B) Note how the coracoid graft restores the pear shape of the glenoid by widening its inferior
diameter.
FIGURE 17. (A) The capsule is repaired to the glenoid by means of the previously placed suture anchors. (B) Completed capsular repair with
graft in place. Note that the coracoid graft is extra-articular.
692
S. S. BURKHART AND J. F. DE BEER
kart repair is not that there is some inherent mechanical inferiority of the arthroscopic repair in comparison with the open repair. The problem is that surgeons
have not developed realistic criteria for patient selection based on mechanical support requirements. Engineers would have done this long ago. Why haven’t
surgeons?
CONCLUSIONS
1.
FIGURE 18. The graft is placed so that it becomes an extraarticular platform that acts as an extension of the articular arc of the
glenoid.
that plastic deformation ever occurs in patients, much
less that it occurs with any regularity.
We are concerned that the current emphasis on
maximizing and even overtightening soft tissue constraints is diverting attention from what we believe is
the real culprit in the failure of arthroscopic Bankart
repairs—the significant bone defect. We must not
penalize patients without bone defects by overconstraining and overtightening their capsular tissues, potentially causing permanent loss of motion, when we
should be focusing our efforts on restoring bone stability to the relatively small subgroup with significant
bone defects.
The problems with the failure of arthroscopic Ban-
2.
3.
Arthroscopic Bankart repairs produce results
equal to open Bankart repairs if there are no
significant structural bone deficits (large engaging Hill-Sachs lesions or large bony Bankart lesions).
Patients with significant bone deficits as defined in this study are not candidates for arthroscopic Bankart repair.
A Hill-Sachs lesion that engages the anterior
glenoid rim in a functional position of combined flexion-abduction– external rotation on
dynamic arthroscopic examination is a contraindication to arthroscopic repair. Such lesions demand either an open capsular shift
procedure (to restrict external rotation and
thereby prevent engagement of the Hill-Sachs
lesion), or else creation of additional articular
arc length by means of a bone graft to the
Hill-Sachs lesion. Rotational osteotomy to
shift the remaining articular cartilage to a
position that effectively creates a more favor-
FIGURE 19. Incorrect placement of coracoid bone graft. (A) The graft must not be placed so that it protrudes laterally to the joint surface
and acts as a bone block. Such placement produces a high incidence of late osteoarthritis. (B) Conversely, the surgeon must also avoid medial
placement of the graft, as this may predispose to recurrent dislocation.
FAILED ARTHROSCOPIC BANKART REPAIR
7.
693
rate is no greater than that of the overall
group. However, contact athletes with bone
deficiency require open surgery aimed at their
specific anatomic deficiencies.
This study does not address HAGL lesions.26
Nonetheless, the surgeon treating contact athletes must be aware that these lesions may be
fairly common in contact athletes.27 Because
there is no bone defect with the HAGL lesion,
arthroscopic repair is reasonable, although
difficult, and our recommendation is for anatomic repair, either open or arthroscopic.
Acknowledgment: The authors acknowledge Chen
Yuan, Ph.D., for his assistance in performing the statistical
analysis for this report.
REFERENCES
FIGURE 20. The upper half of subscapularis is repaired to its
tendon stump. The repaired muscle spans the coracoid bone graft,
and the attached conjoined tendon exits anteriorly between the
upper and lower subscapularis.
able articular arc would be a final possibility,
if all other surgical options had failed.
4. A nonengaging Hill-Sachs lesion, even if it is
large, may be successfully treated by arthroscopic repair of the associated Bankart lesion.
Such lesions may engage in the nonfunctional
position of shoulder extension, or in low
ranges (⬍70°) of shoulder abduction, but will
not engage during functional activities.
5. A bony Bankart lesion that narrows the inferior half of the glenoid to a width that is less
than that of the superior half of the glenoid
(the inverted-pear configuration) is a contraindication to arthroscopic repair. Such lesions
demand a bone graft to the anterior-inferior
glenoid, and our preference is to use the vascularized coracoid graft of the Latarjet procedure. We find the inverted-pear configuration
easier to identify than the 25% glenoid loss
advised by Bigliani et al.22 as a criterion for
coracoid transfer.
6. Rugby players seem to be at risk of having
significant bone deficiencies in association
with anterior dislocation (9 of 21 major bone
deficiencies in this series occurred in rugby
players). We believe that arthroscopic Bankart repairs are reasonable in contact athletes
without bone deficiency because their failure
1. Neer CS, Foster CR. Inferior capsular shift for involuntary
inferior and multidirectional instability of the shoulder. A
preliminary report. J Bone Joint Surg Am 1980;62:897-908.
2. Latarjet M. Techniques chirurgicales dans le traitement de la
luxation anteriointerne recidivante de l’épaule. Lyon Chir
1965;61:313-318.
3. Allain J, Goutallier D, Glorion C. Long-term results of the
Latarjet procedure for the treatment of anterior instability of
the shoulder. J Bone Joint Surg Am 1998;80:841-852.
4. Grana WA, Buckley PD, Yates CK. Arthroscopic Bankart
suture repair. Am J Sports Med 1993;21:348-353.
5. Green MR, Christensen KP. Arthroscopic Bankart procedure:
Two- to five-year followup with clinical correlation to severity
of glenoid labral lesion. Am J Sports Med 1995;23:276-281.
6. Morgan CD, Bodenstab AB. Arthroscopic Bankart suture repair: Technique and early results. Arthroscopy 1987;3:111122.
7. McIntyre LF, Caspari RB. The rationale and technique for
arthroscopic reconstruction of anterior shoulder instability using multiple sutures. Orthop Clin North Am 1993;24:55-58.
8. Bacilla P, Field LD, Savoie FH. Arthroscopic Bankart repair in
a high demand athletic population. Arthroscopy 1997;13:5160.
9. Rowe CR, Patel D, Southmayd WW. The Bankart procedure.
A long-term end-result study. J Bone Joint Surg Am 1978;60:
1-16.
10. Wolf EM. Arthroscopic capsulolabral repair using suture anchors. Orthop Clin North Am 1993;24:59-69.
11. Barber FA, Snyder SJ, Abrams JS, Fanelli GC, Savoie FH III.
Biodegradable suture anchors: Preliminary clinical results. Arthroscopy 1998;14:449 (abstr).
12. Elrod BF. Arthroscopic reconstruction of traumatic anterior
instability. Oper Tech Sports Med 1997;5:215-225.
13. Neviaser TJ. The anterior labroligamentous periosteal sleeve
avulsion lesion: A cause of anterior instability in the shoulder.
Arthroscopy 1993;9:17-21.
14. Thomas SC, Matsen FA. An approach to the repair of avulsion
of the glenohumeral ligaments in the management of traumatic
anterior glenohumeral instability. J Bone Joint Surg Am 1989;
71:506-513.
15. Levine WN, Richmond JC, Donaldson WR. Use of the suture
anchor in open Bankart reconstruction. A follow-up report.
Am J Sports Med 1994;22:723-726.
16. DuToit GT, Roux D. Recurrent dislocation of the shoulder. A
694
17.
18.
19.
20.
21.
S. S. BURKHART AND J. F. DE BEER
twenty four year study of the Johannesburg stapling operation.
J Bone Joint Surg Am 1956;38:1-12.
O’Driscoll SW, Evans DC. Long-term results of staple capsulorrhaphy for anterior instability of the shoulder. J Bone Joint
Surg Am 1993;75:249-258.
Johnson LL. The glenohumeral joint. In: Johnson LL. Diagnostic and surgical arthroscopy of the shoulder. St. Louis:
Mosby, 1993;318-331.
Burkhart SS, Danaceau SM. Articular arc-length mismatch as
a cause of failed Bankart repair. Arthroscopy 2000;16:740744.
Brañes J. Tratamiento de las luxaciones recidivantes de la
articulación escàpulo humeral. Presented at the IV Congress of
Simposia Latino Americana Ortopedica y Traumatologia, Santiago, Chile, 1969.
Weber BG, Simpson LA, Hardegger F. Rotational humeral
osteotomy for recurrent anterior dislocation of the shoulder
associated with a large Hill-Sachs lesion. J Bone Joint Surg
Am 1984;66:1443-1446.
22. Bigliani LU, Newton PM, Steinmann SP, Connor PM,
McIlveen SJ. Glenoid rim lesions associated with recurrent
anterior dislocation of the shoulder. Am J Sports Med
1998;26:41-45.
23. Helfet AJ. Coracoid transplantation for recurring dislocation of
the shoulder. J Bone Joint Surg Br 1958;40:198-202.
24. Baker CL, Uribe JN, Whitman C. Evaluation of acute intial
anterior shoulder dislocations. Am J Sports Med 1990;18:2528.
25. Bigliani LU,Pollock RG, Soslowsky LJ, Flatow EL, Pawluk
RJ, Mow VC. Tensile properties of the inferior glenohumeral
ligament. J Orthop Res 1992;10:187-197.
26. Wolf EM, Cheng JC, Dickson K. Humeral avulsion of the
glenohumeral ligaments as a cause of anterior shoulder instability. Arthroscopy 1995;11:600-607.
27. Bokor DJ, Conboy VB, Olson C. Anterior instability of the
glenohumeral joint with humeral avulsion of the glenohumeral
ligament: A review of 45 cases. J Bone Joint Surg Br 1999;
81:93-96.